[arin-ppml] Creating a market for IPv4 address space inabsenceof routing table entry market

[Tom Vest]

Some confusion is edifying...

On Jun 18, 2008, at 2:54 PM, John Santos wrote:

> On Wed, 18 Jun 2008, David Schwartz wrote:
>
>>
>>>> I don't think anyone who advocates a market in address space
>>>> thinks that implies ownership of address space.  It implies
>>>> ownership of the right to use address space (i.e. a license
>>>> to use that unique set of integers, in the limited context of
>>>> the IPv4 Internet.)
>>
>>> If so, then that would be a MAJOR change to the set of rights wich
>>> currently come with an ARIN allocation. Currently you have a right
>>> to use those integers in devices which support the Internet Protocol
>>> version 4. This holds whether or not you connect the set of devices
>>> to the Internet or not. If you have a need for uniqueness, you can
>>> apply to ARIN and get addresses. Many companies have done so, often
>>> to use in VPNs or private internetworks (also called extranets) in
>>> which companies connect to their trading partners.
>>>
>>> --Michael Dillon
>>
>> Your response is a non sequiter. He said that A implies B. You  
>> replied that
>> would be a major change because we have always felt that C was also  
>> true.
>> However, there is no inconsistency between B and C. In fact, C is
>> independent of B.
>>
>> Perhaps you took his "A implies only B" to mean that he was  
>> claiming your
>> particular superset of B was false. But he did no such thing. He  
>> simply said
>> it wasn't implied by A. However, that takes no position on its  
>> truth or
>> falsity.
>>
>> In this case:
>>
>> A = A market for address space
>> B = Ownership of a right to use address space only on the Internet
>> C = Assignment includes right to use address space in non-Internet  
>> contexts
>>
>> In case it's not clear, he argued that a market in address space  
>> implies
>> that you can own the right to use address space in the limited  
>> context of
>> the IPv4 Internet (A implies or requires just B). He said nothing  
>> about use
>> of address space in any other context. He didn't say you couldn't  
>> get it or
>> didn't have it, he just said that only ownership of a right to use  
>> in the
>> limited context of the Internet is required for a market in address  
>> space.
>>
>> And, in fact, when someone wants to "buy IP addresses", what they  
>> want is
>> the exclusive right to use those IP addresses on the public  
>> Internet. If
>> this right can be had and transferred, then there can be a market  
>> in address
>> space.
>>
>> DS
>
> That's what I meant in my followup, if it was a little incoherent.
>
> Someone else has mail me privately (not sure if it is okay to quote
> him directly) that the dispute with my original statement is not that
> the rights go beyond what I said, but that there are any rights at
> all at stake.  He said there are not, but there is an explicit
> guarantee of uniqueness.  I claim that if this guarantee has any
> meaning, then it implies a right.  Otherwise it is no guarantee
> at all.  Maybe the dispute is over the meaning of the word "right"
> or license.  Or maybe Micheal is claiming that there is a broader
> right than I posited, and my other correspondent is claiming
> narrower rights.
>
> As for extranets, ARIN doesn't grant you the right to use the
> allocated (or assigned) addresses in a non-Internet context,
> you already have the right to use *any* addresses you and
> your trading partners agree to.

> All that ARIN provides in this
> context is a convenient allocation scheme.  However, a typical
> Internet user does not have any explicit contract or agreement
> with the vast majority of other Internet users, so the RIR
> guarantee of uniqueness and exclusivity matters.

Here it sounded like you were asserting something like:

Whatever two parties agree to do bilaterally with respect to  
announcing and accepting any IPv4 address/prefix -- even one that  
might be used elsewhere in a full "public" capacity (i.e., with  
whatever expectation of uniqueness that an official RIR allocation  
implies) is entirely up to the two parties involved; no RIR policies  
or other external requirements or restrictions apply, at least in the  
context of that specific bilateral usage.

However, subsequent elaborations made the claim sound much broader --  
if not quite a declaration of absolute (bilateral) contractual  
precedence over all other restrictions or requirements applicable to  
possession or use of Pv4 addresses.

And that's exactly the problem.

When you buy your PI IPv4 address space from Address Vendor X, are you  
required to participate fully in whois, to keep your whois records  
complete and up-to-date? If the presumed answer is "yes", where does  
the requirement come from? What are the mechanism(s) for ongoing  
verification/accuracy maintenance, and the default response in the  
event of noncompliance? In the event of a loss/failure of accuracy,  
what basis and mechanisms will exist to restore accuracy, or to pursue  
other remedies if that is not possible?

This is a real question, for me anyway, as I am having troubling  
imagining answers other than:

1. No mechanisms other than self-interest; that will be enough to  
guarantee that whois continues to be (or becomes) "good enough"
	-- a faith which is not well supported by the quality of whois in the  
pre-RIR days, back when individual incentives really were the only  
mechanism to preserve accuracy (and the net was small, and largely  
noncommercial, and everyone was chummy).

2. Bilateral address transfer contracts will incorporate whois  
requirements
	-- Just begs the same questions in a different form (where does the  
contractual language requirement come from?), along with many others  
-- e.g., who gets stuck with monitoring/verification, who is harmed  
and may seek redress in the event of a failure, etc.

3. Hierarchical resource certification will solve all of these problems
	-- A plausible answer, but resource certification is a completely  
separate issue, which is not assumed much less required by the  
resource transfer proposals -- and one that is not exactly embraced  
with great enthusiasm by all IPv4 market advocates.

So, we had an interesting if somewhat ambiguous object lesson about  
identifiability tonight:

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&Item=270247528043&Category=11175&_trksid=p3907.m29

The prankster may have intended to illustrate that a transfer market  
is inevitable or unstoppable -- or maybe that a transfer market is  
certain to bring an end to any illusions about the viability of self- 
provisioning/self-governance of address resource identifiability, in  
which case someone will no doubt come along and help us out with that...

Is there another alternative?

TV


  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.arin.net/pipermail/arin-ppml/attachments/20080619/50b8fa36/attachment.htm>